Case Studies / 03
Two forecasting models quantifying how climate change moves energy demand — data contributed to the 2021 IPCC 6th Assessment Report.
Climate policy needs numbers it can defend. Quantifying how a changing climate will move building energy demand across an entire country means modelling at a scale, and a level of transparency, that proprietary tools rarely offer — and a policy body cannot cite what it cannot inspect.
Science and engineering advance fastest when models, data and methods are open and reproducible. Both models were therefore built to be transparent and highly scalable, and released openly as HBLM-USA and DEG-USA, so the method could be examined rather than taken on trust.
Both models were built with deep neural networks and Bayesian statistics, run across one million buildings spanning the entire United States, and published in peer-reviewed journals — the data-driven study and the physics-based study. The resulting demand scenarios were submitted to the IPCC 6th Assessment Report panel, which aggregated them with those of more than 15 other selected teams and published the averaged projections in the report.
Delivered data-driven (HBLM-USA) and physics-based (DEG-USA) forecasts of how climate change will affect energy consumption in the United States to the end of the century.
Contributed data to the IPCC 6th Assessment Report on Climate Change, informing worldwide environmental policy since 2021.